Hi everyone,
I’ve been experimenting with combining traditional feedforward neural networks with memory-augmented architectures like Differentiable Neural Computers to improve short-term trading predictions. My idea is to leverage external memory to capture patterns over longer sequences of market data without losing short-term responsiveness.
Has anyone tried integrating such hybrid models for futures or forex trading? I’m curious about practical performance, training stability, and whether FPGA or GPU acceleration makes a noticeable difference in live environments.
Would love to hear about your experiences, experiments, or resources—especially anything that blends real-time trading with advanced neural network architectures.
I’ve been experimenting with combining traditional feedforward neural networks with memory-augmented architectures like Differentiable Neural Computers to improve short-term trading predictions. My idea is to leverage external memory to capture patterns over longer sequences of market data without losing short-term responsiveness.
Has anyone tried integrating such hybrid models for futures or forex trading? I’m curious about practical performance, training stability, and whether FPGA or GPU acceleration makes a noticeable difference in live environments.
Would love to hear about your experiences, experiments, or resources—especially anything that blends real-time trading with advanced neural network architectures.